Modelling of an Industrial Scale High Pressure Roller Mill ...6.pdf · pressure roller mill named the Poittemill grinding limestone materials for a wide range of parameters (such

MinBaS Område 1 Rapport nr 1:6 Mineral•Ballast•Sten

MinBaS projekt nr 1:4 Optimal malning Delprojekt nr 1,42 Optimal malning - slutrapport

Prediction of Performance of a Commercial

Scale High Pressure Roller Mill (Poittemill) in Production of Limestone Powders

Yanmin Wang1, Eric Forssberg1, Lars Sunnebo2 and Anneli Sköld2

1 Department of Chemical Engineering and Geo-science Luleå University of Technology, S-971 87, Luleå, Sweden 2 Nordkalk AB, S-281 92, Hässleholm, Sweden Luleå Maj 2005

Short summary Experimental data were collected from tests using an industrial scale high pressure roller mill named the Poittemill grinding limestone materials for a wide range of parameters (such as force pressure, circumferential speed of the roll and feed size) at the Ignaberga plant, Nordkalk AB. These data were used to develop models of throughput, size reduction and energy utilisation with the parameters. A performance model with a correction coefficient, which has been developed, can describe the Poittemill throughput at various force pressures and circumferential speed of the rolls. The materials leaked beside the rolls are found to be empirically related to the circumferential speed of the roll in a given force pressure. It is shown that two major parameters such as force pressure and circumferential speed of the roll have an influence on the median size (d50) of the ground product. A coarser product is obtained at a higher circumferential speed of the roll or at a lower force pressure. The force pressure is the most dominant effect on the energy utilisation (ΔP%-0.2 mm/Em) of the ground product. The feed size used has a slight influence on the grinding results. The energy to the Poittemill for comminution is utilised more efficiently at a lower force pressure or a higher circumferential speed of the roll. Empirical models can predict the comminution characteristics with respect to the major parameters in the Poittemill system in dry mode. Product size-energy input relations have been also established, independent of the operating parameters used. Keywords: Model, HPRM, Roller, Limestone, Powder

1

2

Sammanfattning Modell för en högtrycksvalskvarn vid produktion av kalkstensmjöl Försök har genomförts vid malning av kalksten med högtrycksvalskvarn, typ Poitemill i Ignaberga, Nordkalk AB. Vid försöken varierades tryck, varvtal och ingående partikelstorlek. Mätvärdena har använts för att utveckla modeller för kapacitet, nedmalningsgrad och energieffektivitet. Den modell som framtagits kan beskriva genomsättningen för Poitemill vid varierande valstryck och periferihastigheter. Studier gjordes också av det material som läckte vid sidan av valsarna. Resultaten visar att en grövre produkt erhålles vid ett lägre valstryck och vid lägre periferihastighet. Valstrycket har störst betydelse för energieffektiviteten. Den ingående partikelstorleken har liten betydelse. Den tillförda energin utnyttjas mest effektivt vid lågt valstryck eller hög periferihastighet. Sambandet mellan partikelstorlek för produkten och tillförd energi har etablerats och beskrivs i rapporten.

3

4

Contents Page

Introduction 7

Experimental arrangements 8

Materials tested 8

Poittemill system 8

Experimental design and procedure 10

Measurements 11

Prediction of Poittemill throughput 12

1) Modelling of throughput from rolls 12

2) Leakage beside rolls 15

3) Calibration 16

Empirical prediction 17

1) Analysis of variables (ANOVA) 17

2) Relation between median size and parameters applied 19

3) Relation between energy utilisation and parameters applied 21

4) Product size-energy input relation 23

Summary 27

Nomenclature 29

Acknowledgements 31

References 33

Appendix 35

5

6

Introduction

It is known that an efficient comminution with high pressure roller mills occurs

for energy saving and size reduction. One of the roller press mills available is

Poittemill produced by POITTEMILL INGENIERIE Group, France. This mill

combined with air classifiers has been commercially applied into powder

processing plants (Wang and Forssberg, 2004).

To date the modelling of the performance of high pressure roll mill has been

largely restricted to describing size distribution curves from laboratory

machines using the self-similarity principle (Kapur, 1972; Fuerstenau, 1991).

For plant designer and operators interested in how the technology for this type

of mill can be incorporated in comminution circuits. The frequent requirement

is a performance model describing the throughput, size reduction and energy

input of an industrial-scale HPRM used in an in-site processing plant. Model

developments on this modern mill have received some attention in some

countries, particularly in Australia. Recent work has been made on the

performance models, describing the throughput and power draw in addition to

the particle size distribution (Lim and Weller, 1997; Morrell, Tondo and Shi,

1997). Morrell, ET al (1997) has also derived models for the prediction of

power draw, product size distribution and throughput. In order to verify that

these models developed from the lab-scale mills can be used for prediction of

full scale machines, the Julius Kruttschnitt Mineral Research Centre, Australia,

has recently carried out work regarding the HPRM model verification and scale

up (Shi, 2003). In this work, the four rock materials selected to test were from

Rio Tinto, De Beers and BHP Billiton.

Although some researchers have developed models for the purpose in the case

of the HPRM, it is necessary to model the performance of the Poittemill in a

special case of milling a limestone material under corresponding conditions.

The present study will concern modification and development of a performance

model describing the machine throughput and empirical models for relations

between size reduction and energy input under various conditions by a full scale

7

high pressure roller mill (Poittemill) processing limestone materials at the

Ignaberga Plant, Nordkalk AB, Sweden.

Experimental arrangement Material tested

Limestone materials with two feed sizes of >3 mm<40 mm (with < 0.2 mm) and

>3<5 mm (with <0.2 mm), which have been processed at the Ignaberga Plant,

were used in the experiments.

Poittemill system

The grinding tests were conducted in a Pf-RP01 full-scale high-pressure roller

mill (600 mm dia. × 800 mm length rolls) manufactured by POITTEMILL,

France. Figure 1 below shows the Poittemill installed in the Ignaberga Plant

process flowsheet for industrial production of limestone powders. This machine

consists of the two press rolls with their shafts, bearings and bearing housings,

which are installed in a solid frame. The fixed roll rests with its bearing

housings against the frame. The moveable roll is forced by a hydraulic pressure

system against the fixed roll. The rolls are driven through gear boxes. The

grooved lining was used in the high pressure comminution of the limestone

material.

Figure 2 shows the profiled surface of the existing rolls. It can be seen that on

the surface of the existent rolls, along the middle regions appears a curved

profile with varying working gaps due to the wear after long-term use. The mill

is gravity fed, without any external screw devices. The experiments in the

HPRM were performed with limestone as a feed material. The material is

extracted from feed bin into the mill. The circumferential speed can be adjusted

in the range of 0-1.24 m s-1.

8

Figure 1 Poittemill installed in the flowsheet of a processing plant at Ignaberga.

Figure 2 Profiled surface of the rolls

9

Experimental design and procedure

Prediction of throughput: To describe the throughput of the Poittemill by a

performance model, a series of experiments were arranged to run at various

circumferential speeds of rolls (0.48, 0.70, 0.85 and 1.00 m s-1) and force

pressures (3.8 and 5.0 N mm-2). The powdered materials for each condition

were taken from a discharge hopper below the rolls. The leakage material

unground from the sides of the rolls was also examined. The samples for each

condition were taken for 4 to 6 times to ensure the reproducibility. In addition,

all the powdered materials (from the rolls and the leakage) under various

conditions were also taken from the screw channel after de-agglomeration in

order to re-check the total throughput discharged from the mill.

Empirical prediction for size reduction and energy input: A statistical

experimental design of multiple-variable with mixed levels (L18-21×3n) was

employed to perform the dry grinding experiments, as shown in Table 1. The

empirical relations for size reduction and energy input under various conditions

were established through the experiments. The independent variables or

parameters, which affect the performance of the poittemill, were selected for

investigation as follows:

- Feed size, mm;

- Circumferential speed of the rolls, m/s and

- Force pressure, N/mm2.

The responses or dependent variables for the statistical analysis/evaluation of

the experimental results is energy utilisation (the percentage difference of in

particle size below 0.2 mm between the feed and the ground product /energy

input (ΔP%-0.2 mm/Em)) and the median size of the ground product (d50).

The Poittemill system works continuously in dry mode. The powdered material

in the hopper at the top of the system was fed into the grinding rolls. The

material was dry ground to produce different products under various conditions

of the mill such as feed size, circumferential speed of the roll and force

pressure. In each experiment, the powdered samples were taken in the feed pipe

before roll-press grinding and in the product discharge hopper below the rolls

10

for analysis. The samples for particle size analysis under each condition were

taken twice. Also, the reactive power (kW) and the specific energy consumption

(kWh t-1) were determined in each experiment.

Table 1 Experimental design for dry grinding limestone materials by Poittemill system, L18 (21×32)

Test Parameter level no Feed size,

mm Circumferential speed of the roll, m s-1

Force pressure, N mm-2

Energy utilisation or median size

1 1 1 1 R1 2 1 1 2 R2 3 1 1 3 R3 4 1 2 1 R4 5 1 2 2 R5 6 1 2 3 R6 7 1 3 1 R7 8 1 3 2 R8 9 1 3 3 R9 10 2 1 1 R10 11 2 1 2 R11 12 2 1 3 R12 13 2 2 1 R13 14 2 2 2 R14 15 2 2 3 R15 16 2 3 1 R16 17 2 3 2 R17 18 2 3 3 R18 Level 1 2 3 Feed size, mm +3-40 (<0.2) +3-5 (<0.2) Circumferential speed of the roll, m s-1

0.48 0.70 1.00

Force pressure, N mm-2 2.5 3.8 5.0

Measurements

Particle size analysis for the feed material and the ground product was carried

out by sieving method in an Alpine screen. The screen aperture sizes used were

250, 125, 90 and 45 µm. Some press-agglomerates in the sample were de-

agglomerated with a mechanical method prior to the sieving analysis. In this

study, an active AC power (kW) was recorded by taking the power factor into

account with a digital power meter Model 6300 (Kyoritsu Electrical Instruments

11

Works, Ltd., Japan). The net energy consumption of the Poittemill system for

each condition, i.e., the mass specific energy input, Em (kWh t-1), was

determined by , where PMPE mm /= m is the reactive power draft of rolls and

M is the feed rate of the dry solids.

Predication of Poittemill throughput

1) Modelling of throughput from rolls

Based on the roll profile of the Poittemill with various working gaps, the

throughput of the mill can be theoretically described by:

nec QQQ += (1)

where is the mass discharged from both edges region of rolls with a given

working gap (t/h), is the mass discharged from the middle region of the rolls

at a certain working gap (t/h).

eQ

nQ

VLQ eeee ρδ3106.3 ×= (2)

)(106.31

3 VLQ i

n

iiin ρδ∑

=

×= (3)

where Le is the length of the both edges, Li is the length of zone i in the middle

of rolls (m), δe is the working gap between the rolls in both edges regions and δi

is the working gap of zone i in the middle region (m), ρe and ρi are the flake

densities in both edges regions of the rolls and in zone i of the middle of rolls

and in (t/m3), respectively, and V is the circumferential velocity of the rolls

(m/s).

12

Figure 3 shows the experimental data of the Poittemill throughput as well as the

corresponding prediction by Equation (1) at various force pressures and

circumferential speeds of the rolls when a limestone material was dry ground by

the Poittemill system. Clearly, the throughputs measured from the Poittemill

are similar to these predicted by the model at lower circumferential speeds of

the rolls below 1.00 m s-1. However, the predicted throughputs at 1.00 m s-1 are

higher than those measured from the Poittemill grinding the limestone material.

This indicates that the powdered material between the rolls could be slipped at

high speeds (Schoenert, et al. 1990). In other words, Equation (1) cannot be

used to predict the throughput of the mill at high speeds over 1.00 m s-1 in the

case of grinding of the limestone material.

0 20 40 60 80 100 120 140 160 1800

20

40

60

80

100

120

140

160

180

0 20 40 60 80 100 120 140 160 1800

20

40

60

80

100

120

140

160

180

Specific press force: 3.8 N/mm2

0.48 m/s 0.70 m/s 0.85 m/s 1.00 m/s

Specific press force: 5.0 N/mm2

Thro

ughp

ut fr

om ro

lls (m

easu

red)

, t/h

Throughput from rolls (calculated), t/h

Figure 3 Deviation of the throughput from the rolls predicted by

Equation 1 when a limestone material was ground by the Poittemill at various circumferential speeds of roll and force pressures

13

It was indicated that the slip for a certain of material pressed along the roll

surface is a function of the circumferential speed of the rolls (V) and the

dimensionless working gap (δ/D, where δ is a gap between rolls and D is the

roll diameter.). A correction coefficient (f) is suggested to correct the slip effect

and is given by

c

m

QQ

f = (4)

where Qm is the measured throughput from the rolls. Figure 4 shows a plot of

empirical relation between f and Vδ/D.

)/(863.1067.1 DVf δ−= (5)

Equation (1) can be modified with the correction coefficient f. The Poittemill

throughput at various force pressures and circumferential speeds of roll (Qc) can

be modified with f and is theoretically expressed by

ccf QfQ ⋅= (6)

Figure 5 shows the indications of the measured throughput and the predicated

throughput after the correction by Equation (6). Clearly, this model can predict

the throughputs from the rolls of the Poittemill at various force pressures (3.8 to

5.0 N mm-2) and circumferential speeds of roll (0.48 to 1.00 m s-1).

14

0.00 0.01 0.02 0.03 0.04 0.050.5

1.0

1.5

f=1.067-1.863(Vδ/D)

f=Q

m/Q

c

V(δ/D)

Figure 4 Empirical relation between the correction coefficient and V(δ/D)

0 20 40 60 80 100 120 140 160 1800

20

40

60

80

100

120

140

160

180

0 20 40 60 80 100 120 140 160 1800

20

40

60

80

100

120

140

160

180

Specific press force: 5.0 N/mm2

Specific press force: 3.8 N/mm2

0.48 m/s 0.70 m/s 0.85 m/s 1.00 m/s

Thro

ughp

ut fr

om ro

lls (m

easu

red)

, t/h

Throughput from rolls (calculated), t/h

Figure 5 Deviation of the throughput from the rolls predicted by

Equation 6 after the correction when a limestone material was ground by the Poittemill at various circumferential speeds of roll and force pressures

2) Leakage beside rolls

The powdered material leaked beside the rolls occurred when the limestone

material was pressed between rolls in the Poittemill. Figure 6 shows the material

leaked beside the rolls (QmL) as a function of circumferential speed of the rolls

15

(V) at various force pressures. The QL is related to the circumferential speed of

the roll in the case of a certain of force pressure and follows an empirical

relation described by:

)(VBAQL += (7)

where A and B are constants determined by the conditions. It is seen that the

leakage increases with increasing the speed and the force pressure.

0.2 0.4 0.6 0.8 1.0 1.210

15

20

25

30

QL=9.85+15.02(V)

QL=11.45+11.69(V)

3.8 N/mm2

5.0 N/mm2

Leak

age

from

side

s of r

olls

(QL),

t/h

Circumferential velocity of rolls (V), m/s

Figure 6 Material leaked beside the rolls as a function of

circumferential speed of the rolls at various force pressures

3) Calibration

To ensure the respective measurements from the rolls and beside the rolls

(leakage) accurate, all the materials discharged from the Poittemill after de-

agglomeration (QT) were examined. Table 2 gives the results for the

measurements. It can be concluded that the measurements of the materials

discharged from the rolls or the materials leaked beside the rolls are reasonable.

In the table, C** is close to 1, which is better than C*. This indicates that the

prediction by the model for the Poittemill throughput is accurate.

16

Table 2 Measurement of the total throughputs under various conditions for calibration

Force Roll Predicted

Measured

Total

pressure, N/mm2

speed, m/s

From rolls, t/h

Leakage, t/h

Sum, t/h

From rolls, t/h

Leakage, t/h

Sum, t/h

throughput, t/h

C* C**

0.48 66.7 17.1 83.88 66.9 17.6 84.6 82.8 0.97 0.99

3.8 0.70 91.4 19.6 111.1 91.4 18.1 109.5 111.6 1.01 1.00

0.85 114.6 21.4 136.0 116.3 22.1 138.4 136.8 0.99 1.01

1.00 151.9 23.1 175.1 158.7 22.3 181.0 158.4 0.87 0.91

0.48 75.6 17.0 92.6 75.75 19.7 95.45 86.4 0.91 0.93

5.0 0.70 102.3 20.4 122.7 102.1 17.1 119.2 110.8 0.93 0.90

0.85 114.5 22.6 137.1 120.2 24.0 144.2 131.2 0.91 0.96

1.00 165.6 24.6 190.2 167.5 24.6 192.1 183.6 0.96 0.97

C*=QT/(Qm+QmL); C**=QT/(Qc+QL)

Empirical prediction 1) Analysis of variables (ANOVA)

Experiments on three major operating parameters (feed size, circumferential

speed of the roll and force pressure) were carried out. A significant difference

was considered at a confidence level of 95 % (α<0.05), LSD significant

differences test (Montgomery, 1991) in the ANOVA was utilised to analyse a

difference between two sizes of feed, a difference among three levels of

circumferential speed of the roll and a difference among three levels of force

pressure as well. Since the designed experiments were not replicated, it was

assumed that the two-factor and three-factor interaction effects would be

negligible. The ANOVA was used to examine the null hypothesis that the

variations of each independent variable at various levels have no effect on the

energy utilisation index (ΔP %-0.2 mm/Em). The results of the ANOVA in

Table 3(a) demonstrate that the probability for the null hypothesis that the force

pressure has no significant effect is 0.0383. This means that force pressure is the

primary important variable among the parameters studied. The variation of this

parameter dramatically influences the energy utilisation (ΔP %-0.2 mm /Em) in

dry comminution of the coarser limestone materials by the Poittemill. The

probabilities for the feed size and the circumferential speed of the roll are 21.48

% and 21.66 %, respectively. This proves the hypothesis that the variations do

17

not affect the results of the energy utilisation. The effects of feed size and

circumferential speed of the roll have not been demonstrated to affect the

energy utilisation significantly from a statistical viewpoint. This illustrates that

variations of fineness of the ground product appear to be insignificant when the

coarse limestone materials with feed sizes of +3-40 mm (with – 0.2 mm) and

+3-5 mm (with –0.2 mm) were ground at various circumferential speed of the

roll from 0.48 m s-1 to 1.00 m s-1 in the Poittemill system.

Table 3(b) shows that the circumferential speed of the roll and force pressure

are the most dominant effects on the median size (d50) of the ground product

since their F-ratios are 6.40 and 4.90. The results indicate that the

circumferential speed of the roll and the force pressure both are the important

factors, which influence the product fineness obtained in the Poittemill system.

However, the feed size has no effect on the variation of the median size of the

ground product as the probability for this parameter to have an effect for the

null hypothesis is 91.94 %.

Table 3 Results of ANOVA for dry comminution of the coarse limestone by Poittemill system

a) Energy utilisation (ΔP %-0.2 mm /Em)

Source Sum of squares

Df Mean square

F-ratio

Probability

Main parameter F (Feed size, mm) 125,393 1 125.39 1.72 0.2148 V (Speed of the roll, m s-1)

230.593 2 115,29 1.58 0,2466

P (Pressure, N mm-2) 633,373 2 316,68 4.33 0.0383 Residual 877,115 12 73,09 Total (corrected) 1866,474 17

b) Median size (d50)

Source Sum of squares

Df Mean square

F-ratio

Probability

Main parameter F (Feed size, mm) 0,0000067 1 0.0000067 0,01 0.9194 V (Speed of the roll, m s-1)

0,0061818 2 0,0030908 6.40 0,0128

P (Pressure, N mm-2) 0,0080701 2 0,0040351 4.90 0,0277 Residual 0,0075623 12 0,0006302 Total (corrected) 0,0218209 17

All F-ratios are based on the residual mean square error.

18

2) Relation between median size and parameters applied

Figure 7 shows the median size of ground product as a function of

circumferential speed of the roll at various feed sizes and force pressures.

Clearly, as the circumferential speed of the roll is increased, the median size of

ground product shifts to a coarser size, resulting in a decrease of the product

fineness. For instance, a circumferential speed of 0.70 m s-1 produced a median

size of 0.19 mm, while at 1.00 m s-1 a median size of 0.26 mm was obtained at

the feed size of +3-40 mm (with -0.2 mm) for the case of using a force pressure

of 3.8 N mm-2. It is also shown that the product fineness increases with

increasing force pressure at given circumferential speeds of rolls.

0.4 0.6 0.8 1.0 1.20.15

0.20

0.25

0.300.4 0.6 0.8 1.0 1.2

0.15

0.20

0.25

0.30

Feed size: +3-5 mm (with - 0.2 mm)

Feed size: +3-40 mm (with -0.2 mm)

Med

ian

prod

uct s

ize,

d50

, mm

Circumferential speed of roll, m s-1

2.5 N mm-2

3.8 N mm-2

5.0 N mm-2

Figure 7 Median size of ground product as a function of circumferential

speed of the roll at various feed sizes and force pressures

Possibly more micro-cracks are formed in the particles pressed between the rolls

at higher force pressures and lower circumferential speeds of rolls, resulting in a

smaller product size. This is due to the increased compression of the particles at

higher pressures or the prolonged compression of the particles at lower speeds.

It appears that an increased grinding force at higher pressures and lower speeds

19

leads to deeper pressing penetration into the packed particle bed, enabling the

particles to be better gripped between the rolls.

For the purpose of prediction, an empirical model for the estimation of the

performance of dry grinding coarse limestone materials using the Poittemill

system can be established from the results by the statistical approach. The

variables involved were the feed size (F), the force pressure (P) and the

circumferential speed of the roll (V) as well. No interaction effects were

considered in the model due to their insignificant effects. The following

estimated model for the median size of the ground product (d50) could be thus

expressed for dry comminution of the limestone materials in the Poittemill

system.

)(1016.3)(1066.1)(1009.6 224150 VPFkd −−− ×−×−×+=

(Rsq=71.67%) (8)

where k1 is a constant, equalling to 0.293 for the d50 in equation (8). Figure 8

shows the predicted data and the measured results obtained under various

conditions. It is seen that the empirical model may be appropriate for the

predication due to its higher correlation indication. The model is primarily valid

for the ranges of the parameters used for dry grinding of the limestone in this

comminution system.

Figure 8 Predicted median product size and measured data obtained

under various conditions

20

3) Relation between energy utilisation and parameters applied

Figure 9 shows the variations of the energy utilisation (ΔP %-0.2 mm/Em) with

respect to the effects of the force pressure and the circumferential speed of the

roll in the cases of two feed materials. It is apparent that the energy utilisation is

significantly affected by either the force pressure or the circumferential speed of

the roll. The energy input to the Poittemill is utilised more efficiently at a lower

force pressure and a higher circumferential speed of the roll. At a lower speed,

the residence time for the comminution of particles is relatively longer, resulting

in a sufficient probability of forming the more micro-cracks in the particles

pressed between the rolls. Thus, the Poittemill performs more energy efficiently

at a lower circumferential speed of the roll. The mode of the breakage may also

play a role in the efficiency of the Poittemill in this work. At a higher speed,

however, the residence time of particles packed in the mill was much shorter,

resulting in an inefficient breakage in compressive loads in the mill. Since this

latter breakage mechanism at high speeds could be much less efficient than slow

compression, the gentle compression action at a lower speed may have resulted

in a more efficient breakage mechanism.

Insufficient energy utilisation for the rolls pressing the particles in the mill with

a high force pressure over 3.8 N mm-2 occur in the range of circumferential

speeds of roll (0.48-1.00 m s-1), particularly for a coarser limestone material

with a feed size of +3-40 mm (with –0.2 mm). The energy losses occur in an

inefficient friction/compression of the rolls rather than an efficient comminution

of the particles. However, if the particles between the rolls that are packed under

an active compressive load in the mill are nearly optimum, the energy can be

efficiently utilised for the grinding action at a given circumferential speed.

Figure 9 also shows that an optimum energy utilisation in the case of a

circumferential speed of 0.7 m s-1 may occur when the limestone materials were

dry ground at higher force pressures.

Empirical predication of a relation between energy utilisation and parameters

used for dry grinding coarse limestone materials using the Poittemill system can

21

be established from the results by the statistical approach. The variables

involved were the feed size (F), the force pressure (P), and the circumferential

speed of the roll (V) as well. No interaction effects were considered in the

model due to their insignificant effects. The following estimated model for the

energy utilisation against these parameters used could be thus expressed for dry

comminution of the limestone materials in the Poittemill system.

)(462.7)(439.3)(158.0/)2.0(% 2 VPFkEmmP m −−−=−Δ

(Rsq=68.91%) (9)

where k2 is a constant, equalling to 37.81 for the energy utilisation in equation

(9). Figure 10 shows the predicted data and the measured results obtained under

various conditions. It is seen that the empirical model may be appropriate for

the predication due to its fair correlation indication. The model is primarily

valid for the ranges of the parameters used for dry grinding of the limestone in

this Poittemill system.

0.4 0.6 0.8 1.0 1.20

5

10

15

20

25

300.4 0.6 0.8 1.0 1.2

0

5

10

15

20

25

30

35

40

Feed size: +3-5 mm (with - 0.2 mm)

Feed size: +3-40 mm (with -0.2 mm)

Ener

gy u

tiliz

atio

n, (%

-0.2

mm

)(kW

h)-1t

Circumferential speed of roll, m s-1

2.5 N mm-2

3.8 N mm-2

5.0 N mm-2

Figure 9 Energy utilisation (ΔP %-0.2 mm/Em) with respect to effects of

the force pressure and the circumferential speed of the roll in the cases of two feed materials

22

Figure 10 Predicted energy utilisation and measured results obtained

under various conditions 4) Product size-energy input relation

The specific energy consumption of the Poittemill with the corresponding size

reduction is the main concern for processing. The important implication for the

nature of the particle size distributions is that the size spectrum (see Appendix)

is driven forward on its trajectory solely by reduction in one parameter, namely,

median size (d50) or the percentage difference between the feed and the product

below 0.2 mm. A model that ensures that the energy for size reduction is

compatible with that provided by the motor should contain a relation between

product size and energy input. Also, the prediction of comminution behaviour

in the Poittemill system and the subsequent development of scale-up design

relationships require a mathematical model relating energy input to the resultant

product fineness. The product median size (d50) and the percentage difference in

the particle size below 0.2 mm between the feed and the ground product (ΔP %-

0.2 mm) respectively are plotted as a function of specific energy input in

Figures 11 and 12. The straight lines in the figures were drawn by using the

following equations:

mEd 128.0331.050 −= (10)

32.313.24)2.0(% −=−Δ mEmmP (11)

23

with the correlation coefficients of 0.84 and 0.71. The product median size or

the percentage difference in the particle size below 0.2 mm varies linearly with

the specific grinding energy expended in the Poittemill system. The fact that all

the data obtained by dry comminution of limestone with the Poittemill are

relatively well described by a single straight line is important for scale-up. It is

possible to predict the fineness of the ground product (d50 or ΔP %-0.2 mm) for

a given energy input, independent of the operating parameters used.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00.0

0.1

0.2

0.3

0.4

0.5

Experimental data_____ Prediction

Med

ian

prod

uct s

ize,

d50

, mm

Specific energy input, kWh t-1

Figure 11 Median size of product (d50) as a function of specific energy input

24

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00

10

20

30

40

50

60

Experimental data_____ Prediction

ΔP %

-0.2

mm

Specific energy input, kWh t-1

Figure 12 Percentage difference in the particle size below 0.2 mm

between the feed and the ground product (ΔP %-0.2 mm) as a function of specific energy input

25

26

Summary

A performance model for throughput of an industrial scale high pressure roller

mill named the Poittemill has been developed for machine milling of limestone

materials in the processing plant (Nordkalk AB) located at Ignaberga, Sweden.

In the model, a correction coefficient has been applied to predict the maximum

throughput of the Poittemill at a high speed. The materials leaked beside the

rolls increase with increasing the circumferential speed of the roll and the force

pressure, obtaining the corresponding empirical relation at a given force

pressure.

Experimental results indicate that the main factors influencing the median size

(d50) of the ground product in a decreasing order of importance are force

pressure and circumferential speed of the roll. The force pressure is the most

dominant effect on the energy utilisation (ΔP%-0.2 mm/Em) of the ground

product. The feed size has a slight influence on the grinding results. An increase

of circumferential speed of the roll or a decrease of force pressure results in a

coarser ground product. The energy to the Poittemill for dry comminution of

limestone materials is utilised more efficiently at a lower force pressure or a

higher circumferential speed of the roll. Empirical models can be used to predict

the comminution characteristics with respect to the major parameters in the

Poittemill system in dry mode. Product size-energy input relations have been

also established, independent of the operating parameters used.

27

28

Nomenclature

A - constant

B - constant

D – roll diameter, m

Em – specific energy consumption, kWh/t

F – Feed size

f – correction coefficient

Le - length of the edges regions of rolls at a given working gap, m

Li – length of zone i in the middle region of rolls at a certain working gap, m

M - feed rate of the dry solids, t/h

P – force pressure, N/mm2

Pm - reactive power draft of rolls, kW

Qc – predicted throughput from the rolls, t/h

Qcf - predicted throughput from rolls modified with a correction coefficient, t/h

Qe - throughput from the edges regions of rolls at a given working gap, t/h

QL – leaked materials beside the rolls, t/h

Qm – measured throughput from the rolls, t/h

Qn – throughput from the middle region of rolls at a certain working gap, t/h

QT – total throughput discharged from the mill, t/h

V – circumferential speed of the roll, m/s

ρe – flake density in the edges region of rolls at a given working gap, t/m3

ρi - flake density in zone i of the middle region of rolls at a certain working gap,

t/m3

δe – working gap between the rolls in the edges regions, m

δi – working gap between the rolls in zone i of the middle region, m

29

30

Acknowledgements

The authors wish to thank the MinBaS and the Swedish Mineral Processing

Research Association (MinFo) for supporting this work. Many thanks go to the

Ignaberga Plant, Nordkalk AB, for providing the Poittemill system and the

corresponding facilities. The authors are also grateful to Mrs Anneli Sköld, Mr

Roger Nilsonius and Mr Bertil Persson at the Ignaberga plant for arrangement

and assistance of the experiments in this work.

31

32

References

Fuerstenau, D. W., Shukla, A, Kapur, P. C., Energy consumption an product

size distributions in choke-fed high compression roll mills, Inter. J. of Miner.

Process, 32, 1991, pp. 59-79.

Kapur, P. C., 1972, Self-preserving size spectra of comminuted particles, Chem.

Engng. Sci., 27, 1972, pp. 425-431.

Lim, W. L. and Weller, K. R., Modelling of throughput in the high pressure

grinding rolls, Proceeding of the XX IMPC, Roller mills, Vol. 2, Aachen, 1997,

pp. 173-184.

Morrell, S., Shi, F. and Tondo, L., Modelling and scale-up of high pressure

grinding rolls, Proceeding of the XX IMPC, Roller mills, Vol. 2, Aachen, 1997,

pp. 129-140.

Montgomery, D. Design and analysis of experiments, 3rd edition, John Wiley &

Sons, New York, 1991.

Shi, F., Personal communication, in JKMRC, Australia, 2003.

Schoenert, K. and Lubjuhn, U., Throughput of high compression roller mills

with plane and corrugated rolls, 7th European Symposium on Comminution,

Ljubljana, 1990, pp. 747-763.

Wang, Y. and Forssberg, E., International overview and outlook on

comminution technology, GHRR, Gällivare, Feb. 2004

33

34

Appendix

Particle size distributions of all samples of the feed materials and the

ground products from the Poittemill system

35

36